Organized Continuity Panel Reassignment

Having a continuity panel, with a distribution of patients by sex, disease, and sociodemographic factors, has been identified as valuable for residency training.1,2 The Residency Review Committees for Family Medicine, Internal Medicine, Pediatrics, and Internal Medicine–Pediatrics require residents to care for a defined group of patients in continuity clinics but do not define the ideal patient mix or panel size beyond requiring “variety” reflective of the specialty and “adequate” sex and age diversity.3–6 For many programs this requirement means building continuity panels for incoming residents by reassigning patients from graduating residents. There are several challenges in this process, yet there is scant literature describing the annual panel-reassignment process beyond discussing panel size.7,8 Reassignment processes vary across programs from seemingly random processes to those that seem deliberate and organized.

Residents depend on their continuity clinics to develop competencies for ambulatory management of patients. Continuity panels that vary significantly by age, race/ethnicity, sex, and types of diagnosis provide a less-equivalent educational experience. The growing emphasis on the patient-centered medical home9 highlights the need for more equivalent patient-care experiences to support training in population-based care.10,11 

The Palmetto Health Family Medicine Residency, affiliated with the University of South Carolina School of Medicine's Department of Family and Preventive Medicine (DFPM), concurrently trains 30 residents in family medicine. Historically, panel reassignment was overseen by the department's Quality Improvement Committee. This process moved patients from the panels of residents leaving the practice to incoming postgraduate year-1 (PGY-1) and rising PGY-2 and PGY-3 residents. The reassignments were not done using a preplanned methodology, were time consuming, often resulted in family members being assigned to different providers, and created panels with widely variable demographics.

Our primary goal was to use a more structured approach to panel reassignment to create balanced panels for resident physicians. We also wanted to enhance residency curricula around patient panel management, identify key indicators for successful panel reassignment, develop objective measures of reassignment outcomes, and assess the effect on these measures.

We instituted the new reassignment process for the 2007–2008 academic year. A preimplementation survey of faculty, residents, and staff indicated a desire for more varied diagnoses; protecting family units; evenly distributing patients requiring special attention (eg, those with chronic illness or high use); a larger proportion of young adults, women, and children; and a de-emphasis on insurance status.

We used this information to create a reassignment process that used race/ethnicity, age, sex, diabetes status, and the number of previous visits as key factors for panel balancing. Targets for each factor were set according to its overall distribution in the practice. The new process began by identifying current panel assignments via an electronic medical record (EMR) database query using Crystal Reports (SAP, Newton Square, PA). Information (patient name, date of birth, sex, race/ethnicity, diabetes status, and number of visits in the previous 36 months) was exported to a Microsoft Excel (Redmond, WA) spreadsheet. A summary worksheet aggregated the current number of patients in each panel by the characteristics of interest. We were able to identify how many patients, by characteristics, each provider needed for balance.

We set a goal to balance resident panels beginning with the 2007–2008 academic year (table 1). Patients were moved from the graduating resident panels to the panels of PGY-1s and rising residents based on the number and types of patients needed. This process initially proved to be time consuming. For the 2008–2009 academic year, a custom JavaScript program was created to automate this process, reducing the time needed to reassign panels to less than 4 hours.

The patient distributions within each panel (PGY-1 2005, PGY-1 2006, PGY-1 2007, PGY-1 2008, and PGY-1 2009) were tested using Wald χ² tests to detect significant differences.

Balanced panels presented an opportunity to tailor the curriculum, beginning with PGY-1 residents receiving their panels on the first day of residency. These panels were used as a basis for instruction on the principles of quality improvement, evidence-based guidelines for diabetes care, proactive care, health maintenance, and establishing continuity visits with selected patients. In addition, the newly balanced panels were used in existing rotations. During their community medicine rotation, PGY-1 residents studied their panels to identify environmental and contextual factors that would influence health, health behaviors, and health care–seeking behaviors. During the pharmacology rotation, residents reviewed their patients with diabetes to assess compliance with National Committee for Quality Assurance (NCQA)12 standards and participated in clinical diabetic education visits with patients from their own panel.

Following implementation, we investigated the effect that panel balancing had on care quality using 2 NCQA process measures for diabetes care: the receipt of 2 or more hemoglobin A1c measures in the past 12 months and the receipt of at least 1 lipid panel performed in the past 12 months.12 We used data from patients assigned to PGY-1 residents in the 2005–2006 (period 1), 2006–2007 (period 2), and 2007–2008 academic years (period 3), as well as faculty physicians from 2005–2008.

Analysis began with the estimation of the percentage of patients in a provider's panel that met each quality indicator (ie, faculty, PGY-1 2005, PGY-1 2006, and PGY-1 2007), at the beginning and end of each academic year. Differences between the percentage estimates were assessed for each provider group, using paired Student t tests.

The project was approved by the Palmetto Health Institutional Review Board and the University of South Carolina Institutional Review Board.

Compared with previous years, the PGY-1 2007–2009 classes had balanced panels that reflected the distribution of the patient population within the FMC (table 1). All PGY-1 residents were assigned between 74 and 77 patients, with equivalent numbers of women, minorities, elderly, patients with diabetes, and a small number of patients with high health use. Comparatively, the PGY-1 distributions seen in 2005 and 2006 had much more variation in the number of patients with diabetes, race/ethnicity distributions, and those under age 21 years.

Each physician classification group (PGY-1, PGY-2, PGY-3, faculty) demonstrated an improvement in the quality indicators for each period (table 2). For the hemoglobin A1c quality indicator, large increases in patients meeting appropriate follow-up frequencies and testing were seen in the first year of practice for both the PGY-1 2005 and 2007 classes. The PGY-1 2005 class, however, did not maintain this increase in their second year but saw an additional improvement in their third year, which was similar to their first year and the first year of the PGY-1 2007 class. The PGY-1 2006 class maintained a large improvement in the process measure in their second year. These significant improvements during period 3 for the resident classes correspond with the balanced panel process, particularly when compared with the smaller faculty improvement during the period. The same trends, however, were not seen for the cholesterol measure. All 3 groups had large improvements in lipid panel measurements in their first years but were unable to maintain this improvement in subsequent years (table 2).

We demonstrated that it is feasible to perform a large, structured patient reassignment process in a residency program leading to balanced panels. These balanced panels, combined with an integrated curriculum, aid residents in obtaining a uniform and comprehensive patient care and educational experience. Residents and faculty indicated a high level of satisfaction with the process. The PGY-1 residents were able to spend time early in residency analyzing their panels, identifying diabetes care needs, and addressing those needs during continuity clinics and group visits. The analysis of quality indicators for diabetes care suggests a trend toward improvement for residents with balanced panels in their first year and for subsequent years, although the magnitude of these improvements varied. For the cholesterol measure, improvement was seen only in the first year and was not sustained in later years of training.

This analysis has several limitations. We were not able to isolate the effect of the balanced panel on quality measures and on other factors that would affect care delivery, such as quality improvement activities and ongoing practice redesign. In addition, variation in residents' baseline clinical knowledge and skills, and care philosophies may have altered their approach to care.

Overall, the balanced-panel project outcome appears to be a successful system-level enhancement for improving family medicine residents' curricular and clinical experience. The panels also promoted parity in patient-care experience, a goal consistent with the Accreditation Council for Graduate Medical Education Outcomes Project.13 Future analysis will determine whether residents with balanced panels for their entire residency were more successful in reaching accepted chronic disease quality indicators.